Hi Lee, On 24 October 2017 11:25, Lee Jones wrote: > Subject: Re: [PATCH 2/6] mfd: da9062: make register CONFIG_I writable > > On Tue, 17 Oct 2017, Michael Grzeschik wrote: > > From: Stefan Christ <s.christ@xxxxxxxxx> > > > > Make the config register CONFIG_I writable to change the watchdog mode. > > The datasheet does not mention exactly whether these registers should be > > read-only or writeable. But writing to those registers works at runtime. > > Only on reset they are reseted. > > > > Signed-off-by: Stefan Christ <s.christ@xxxxxxxxx> > > Signed-off-by: Christian Hemp <c.hemp@xxxxxxxxx> > > --- > > drivers/mfd/da9062-core.c | 4 ++++ > > 1 file changed, 4 insertions(+) > > For my own reference: > Acked-for-MFD-by: Lee Jones <lee.jones@xxxxxxxxxx> Setting CONFIG_I to be writable (and allowing WATCHDOG_SD bit changes to that register) is probably not a great idea. On the face of it, the change is fairly innocuous but it will allow the PMIC's behaviour to be modified during kernel run-time. That change can be done, but making it a permanent change in the general Linux kernel driver would indicate it is okay to change the CONFIG_I registers during normal run-time operation, and generally I do not think this is okay. - [PATCH 3/6] watchdog: da9062: reset board on watchdog timeout https://www.spinics.net/lists/linux-watchdog/msg12680.html I suggested if CONFIG_I is to be altered, then this register could be modified before the Linux kernel driver has started, e.g. early in the bootloader; or the change chould be done in hardware by writing the persistent OTP in the device. If necessary, the Linux kernel driver can read the CONFIG_I register and modify its behaviour accordingly. In both cases, a write access change to CONFIG_I would not be needed. Regards, Steve ��.n��������+%������w��{.n�����{���rh���ܨ}���Ơz�j:+v�����w����ޙ��&�)ߡ�a����z�ޗ���ݢj��w�f
